Dr. Art: Diabetes Re-Admission Risk Tool

Project Details

My RoleData Scientist
SkillsMachine Learning, Data Visualization, Python
Innovative machine learning and data visualization techniques are used to analyze and communicate the risk of early readmission for hospitalized diabetes patients. Dr. Art was a capstone project for the UC Berkeley Master of Information and Data Science degree, and earned a 1st Place Hal R. Varian MIDS Capstone Award.
graphed results

Machine Learning

Training machine learning models on hundreds of thousands of real diabetes-related hospital admissions yielded predictive power that significantly exceeded the industry standard algorithmic approach (LACE).

decision tree path visualization

Innovative Visualization

One of Dr. Art's groundbreaking features was the use of visualization to communicate not just the result of machine learning analysis, but the actual path generated by a decision tree model. This allows healthcare professionals to see why a patient's risk is high or low, not just a number without context.